About

My studies broadly align with what's known as computational social science. That said, a more accurate description would be data science applied to the study of social dynamics and emergent behavior online.

My main research interests include misinformation spreading, online polarization, coordinated behavior, and the societal impact of large language models (LLMs). I take a data-driven approach using mostly tools from network science, along with statistics and machine learning.

I hold an M.Sc. in Data Science and a B.Sc. in Physics. Currently, I'm a Ph.D. candidate under the supervision of Prof. Walter Quattrociocchi at CDCS (Sapienza University of Rome).

Publications and preprints

• "Ideology and polarization set the agenda on social media" , Scientific Reports, 2025

  authors:
  Edoardo Loru, Alessandro Galeazzi, Anita Bonetti, Emanuele Sangiorgio, Niccolò Di Marco, Matteo Cinelli, Max Falkenberg, Andrea Baronchelli, Walter Quattrociocchi
  
  abstract:
  The abundance of information on social media has reshaped public discussions, shifting attention to the mechanisms that drive online discourse. This study analyzes large-scale Twitter (now X) data from three global debates—Climate Change, COVID-19, and the Russo-Ukrainian War—to investigate the structural dynamics of engagement. Our findings reveal that discussions are not primarily shaped by specific categories of actors, such as media or activists, but by shared ideological alignment. Users consistently form polarized communities, where their ideological stance in one debate predicts their positions in others. This polarization transcends individual topics, reflecting a broader pattern of ideological divides. Furthermore, the influence of individual actors within these communities appears secondary to the reinforcing effects of selective exposure and shared narratives. Overall, our results underscore that ideological alignment, rather than actor prominence, plays a central role in structuring online discourse and shaping the spread of information in polarized environments.
  
  doi:
  10.1038/s41598-025-19776-z
  
  cite:
  Loru, E., Galeazzi, A., Bonetti, A., Sangiorgio, E., Di Marco, N., Cinelli, M., Falkenberg, M., Baronchelli, A., & Quattrociocchi, W. (2025). Ideology and polarization set the agenda on social media. Scientific Reports, 15(1). https://doi.org/10.1038/s41598-025-19776-z
  {bib}

• "The simulation of judgment in LLMs" , PNAS, 2025

  authors:
  Edoardo Loru, Jacopo Nudo, Niccolò Di Marco, Alessandro Santirocchi, Roberto Atzeni, Matteo Cinelli, Vincenzo Cestari, Clelia Rossi-Arnaud, Walter Quattrociocchi
  
  abstract:
  Large Language Models (LLMs) are increasingly embedded in evaluative processes, from information filtering to assessing and addressing knowledge gaps through explanation and credibility judgments. This raises the need to examine how such evaluations are built, what assumptions they rely on, and how their strategies diverge from those of humans. We benchmark six LLMs against expert ratings—NewsGuard and Media Bias/Fact Check—and against human judgments collected through a controlled experiment. We use news domains purely as a controlled benchmark for evaluative tasks, focusing on the underlying mechanisms rather than on news classification per se. To enable direct comparison, we implement a structured agentic framework in which both models and nonexpert participants follow the same evaluation procedure: selecting criteria, retrieving content, and producing justifications. Despite output alignment, our findings show consistent differences in the observable criteria guiding model evaluations, suggesting that lexical associations and statistical priors could influence evaluations in ways that differ from contextual reasoning. This reliance is associated with systematic effects: political asymmetries and a tendency to confuse linguistic form with epistemic reliability—a dynamic we term epistemia, the illusion of knowledge that emerges when surface plausibility replaces verification. Indeed, delegating judgment to such systems may affect the heuristics underlying evaluative processes, suggesting a shift from normative reasoning toward pattern-based approximation and raising open questions about the role of LLMs in evaluative processes.
  
  doi:
  10.1073/pnas.2518443122
  
  cite:
  Loru, E., Nudo, J., Di Marco, N., Santirocchi, A., Atzeni, R., Cinelli, M., Cestari, V., Rossi-Arnaud, C., & Quattrociocchi, W. (2025). The simulation of judgment in LLMs. Proceedings of the National Academy of Sciences, 122(42). https://doi.org/10.1073/pnas.2518443122
  {bib}

• "Patterns, Models, and Challenges in Online Social Media: A Survey" , arXiv, 2025

  authors:
  Niccolò Di Marco, Anita Bonetti, Edoardo Di Martino, Edoardo Loru, Jacopo Nudo, Mario Edoardo Pandolfo, Giulio Pecile, Emanuele Sangiorgio, Irene Scalco, Simon Zollo, Matteo Cinelli, Fabiana Zollo, Walter Quattrociocchi
  
  abstract:
  The rise of digital platforms has enabled the large scale observation of individual and collective behavior through high resolution interaction data. This development has opened new analytical pathways for investigating how information circulates, how opinions evolve, and how coordination emerges in online environments. Yet despite a growing body of research, the field remains fragmented and marked by methodological heterogeneity, limited model validation, and weak integration across domains. This survey offers a systematic synthesis of empirical findings and formal models. We examine platform-level regularities, assess the methodological architectures that generate them, and evaluate the extent to which current modeling frameworks account for observed dynamics. The goal is to consolidate a shared empirical baseline and clarify the structural constraints that shape inference in this domain, laying the groundwork for more robust, comparable, and actionable analyses of online social systems.
  
  doi:
  10.48550/arXiv.2507.13379
  
  cite:
  Di Marco, N., Bonetti, A., Di Martino, E., Loru, E., Nudo, J., Pandolfo, M. E., Pecile, G., Sangiorgio, E., Scalco, I., Zollo, S., Cinelli, M., Zollo, F., & Quattrociocchi, W. (2025). Patterns, Models, and Challenges in Online Social Media: A Survey (Version 1). arXiv. https://doi.org/10.48550/ARXIV.2507.13379
  {bib}

• "Generative Exaggeration in LLM Social Agents: Consistency, Bias, and Toxicity" , arXiv, 2025

  authors:
  Jacopo Nudo, Mario Edoardo Pandolfo, Edoardo Loru, Mattia Samory, Matteo Cinelli, Walter Quattrociocchi
  
  abstract:
  We investigate how Large Language Models (LLMs) behave when simulating political discourse on social media. Leveraging 21 million interactions on X during the 2024 U.S. presidential election, we construct LLM agents based on 1,186 real users, prompting them to reply to politically salient tweets under controlled conditions. Agents are initialized either with minimal ideological cues (Zero Shot) or recent tweet history (Few Shot), allowing one-to-one comparisons with human replies. We evaluate three model families (Gemini, Mistral, and DeepSeek) across linguistic style, ideological consistency, and toxicity. We find that richer contextualization improves internal consistency but also amplifies polarization, stylized signals, and harmful language. We observe an emergent distortion that we call "generation exaggeration": a systematic amplification of salient traits beyond empirical baselines. Our analysis shows that LLMs do not emulate users, they reconstruct them. Their outputs, indeed, reflect internal optimization dynamics more than observed behavior, introducing structural biases that compromise their reliability as social proxies. This challenges their use in content moderation, deliberative simulations, and policy modeling.
  
  doi:
  10.48550/arXiv.2507.00657
  
  cite:
  Nudo, J., Pandolfo, M. E., Loru, E., Samory, M., Cinelli, M., & Quattrociocchi, W. (2025). Generative Exaggeration in LLM Social Agents: Consistency, Bias, and Toxicity (Version 1). arXiv. https://doi.org/10.48550/ARXIV.2507.00657
  {bib}

• "Modelling Infodemics on a Global Scale: A 30 Countries Study using Epidemiological and Social Listening Data" , arXiv, 2025

  authors:
  Edoardo Loru, Marco Delmastro, Francesco Gesualdo, Matteo Cinelli
  
  abstract:
  Infodemics are a threat to public health, arising from multiple interacting phenomena occurring both online and offline. The continuous feedback loops between the digital information ecosystem and offline contingencies make infodemics particularly challenging to define operationally, measure, and eventually model in quantitative terms. In this study, we present evidence of the effect of various epidemic-related variables on the dynamics of infodemics, using a robust modelling framework applied to data from 30 countries across diverse income groups. We use WHO COVID-19 surveillance data on new cases and deaths, vaccination data from the Oxford COVID-19 Government Response Tracker, infodemic data (volume of public conversations and social media content) from the WHO EARS platform, and Google Trends data to represent information demand. Our findings show that new deaths are the strongest predictor of the infodemic, measured as new document production including social media content and public conversations, and that the epidemic burden in neighbouring countries appears to have a greater impact on document production than the domestic one. Building on these results, we propose a taxonomy that highlights country-specific discrepancies between the evolution of the infodemic and the epidemic. Further, an analysis of the temporal evolution of the relationship between the two phenomena quantifies how much the discussions around vaccine rollouts may have shaped the development of the infodemic. The insights from our quantitative model contribute to advancing infodemic research, highlighting the importance of a holistic approach integrating both online and offline dimensions.
  
  doi:
  10.48550/arXiv.2501.19016
  
  cite:
  Loru, E., Delmastro, M., Gesualdo, F., & Cinelli, M. (2025). Modelling Infodemics on a Global Scale: A 30 Countries Study using Epidemiological and Social Listening Data (Version 1). arXiv. https://doi.org/10.48550/ARXIV.2501.19016
  {bib}

• "Decoding Musical Evolution Through Network Science" , arXiv, 2025

  authors:
  Niccolò Di Marco, Edoardo Loru, Alessandro Galeazzi, Matteo Cinelli, Walter Quattrociocchi
  
  abstract:
  Music has always been central to human culture, reflecting and shaping traditions, emotions, and societal changes. Technological advancements have transformed how music is created and consumed, influencing tastes and the music itself. In this study, we use Network Science to analyze musical complexity. Drawing on ≈20,000 MIDI files across six macro-genres spanning nearly four centuries, we represent each composition as a weighted directed network to study its structural properties. Our results show that Classical and Jazz compositions have higher complexity and melodic diversity than recently developed genres. However, a temporal analysis reveals a trend toward simplification, with even Classical and Jazz nearing the complexity levels of modern genres. This study highlights how digital tools and streaming platforms shape musical evolution, fostering new genres while driving homogenization and simplicity.
  
  doi:
  10.48550/arXiv.2501.07557
  
  cite:
  Di Marco, N., Loru, E., Galeazzi, A., Cinelli, M., & Quattrociocchi, W. (2025). Decoding Musical Evolution Through Network Science (Version 1). arXiv. https://doi.org/10.48550/ARXIV.2501.07557
  {bib}

• "Patterns of linguistic simplification on social media platforms over time" , PNAS, 2024

  authors:
  Niccolò Di Marco, Edoardo Loru, Anita Bonetti, Alessandra Olga Grazia Serra, Matteo Cinelli, Walter Quattrociocchi
  
  abstract:
  Understanding the impact of digital platforms on user behavior presents foundational challenges, including issues related to polarization, misinformation dynamics, and variation in news consumption. Comparative analyses across platforms and over different years can provide critical insights into these phenomena. This study investigates the linguistic characteristics of user comments over 34 y, focusing on their complexity and temporal shifts. Using a dataset of approximately 300 million English comments from eight diverse platforms and topics, we examine user communications’ vocabulary size and linguistic richness and their evolution over time. Our findings reveal consistent patterns of complexity across social media platforms and topics, characterized by a nearly universal reduction in text length, diminished lexical richness, and decreased repetitiveness. Despite these trends, users consistently introduce new words into their comments at a nearly constant rate. This analysis underscores that platforms only partially influence the complexity of user comments but, instead, it reflects a broader pattern of linguistic change driven by social triggers, suggesting intrinsic tendencies in users’ online interactions comparable to historically recognized linguistic hybridization and contamination processes.
  
  doi:
  10.1073/pnas.2412105121
  
  cite:
  Di Marco, N., Loru, E., Bonetti, A., Serra, A. O. G., Cinelli, M., & Quattrociocchi, W. (2024). Patterns of linguistic simplification on social media platforms over time. Proceedings of the National Academy of Sciences, 121(50). https://doi.org/10.1073/pnas.2412105121
  {bib}

• "The influence of coordinated behavior on toxicity" , Online Social Networks and Media, 2024

  authors:
  Edoardo Loru, Matteo Cinelli, Maurizio Tesconi, Walter Quattrociocchi
  
  abstract:
  In the intricate landscape of social media, genuine content dissemination may be altered by a number of threats. Coordinated Behavior (CB), defined as orchestrated efforts by entities to deceive or mislead users about their identity and intentions, emerges as a tactic to exploit or manipulate online discourse. This study delves into the relationship between CB and toxic conversation on X (formerly known as Twitter). Using a dataset of 11 million tweets from 1 million users preceding the 2019 UK general election, we show that users displaying CB typically disseminate less harmful content, irrespective of political affiliation. However, distinct toxicity patterns emerge among different coordinated cohorts. Compared to their non-CB counterparts, CB participants show marginally higher toxicity levels only when considering their original posts. We further show the effects of CB-driven toxic content on non-CB users, gauging its impact based on political leanings. Our findings suggest that CB only has a limited impact on the toxicity of digital discourse.
  
  doi:
  10.1016/j.osnem.2024.100289
  
  cite:
  Loru, E., Cinelli, M., Tesconi, M., & Quattrociocchi, W. (2024). The influence of coordinated behavior on toxicity. Online Social Networks and Media, 43–44, 100289. https://doi.org/10.1016/j.osnem.2024.100289
  {bib}

---

Conferences

• Contributed talk at IC²S² 2025, "Decoding AI Judgment: How LLMs Assess News Credibility and Bias"
• Contributed talk at NetSci 2025, "A Compression-Based Approach to Detecting Automated and Coordinated Behavior on Social Media"

Coding and plotting and...

I conduct all my analyses using R (mostly) or Python. Lately, I've also been getting into Rust, thanks to its promises of performance and safety as well as modern tooling and syntax. So far, it's lived up to the expectations, though the syntax can get a bit cumbersome.

Here's what I rely on in my day-to-day work:

• data manipulation: polars, dplyr, tidyr, data.table
• data viz: ggplot2, plotnine, gephi
• llms: ellmer, adk, ollama
• other pkgs: igraph, cli, httr2, graph-tool, markitdown
• aux tools: docker, uv, renv, micromamba